Research On The Neural-Network-based Pattern Recognition To The Wheeled Vehicle

It is meaningful and practical to study mobile manipulators which have the capability of image processing and pattern recognition. Lots of cases of manipulator vehicle have been researched at home and abroad. In application, manipulator vehicle with vision ability not only can rescue people in dangerous area, it also can care for elderly people and handicapped.The paper is based on the ART neural network and pattern recognition for the manipulator vehicle. In the recent years, with the development of intelligent control technology, it has been used widely in automation and electronic field. Moreover, the intelligent vehicle is the best. As it can reflect the advantage of intelligent control, the vehicle is used as the object of intelligent control. Theoretically, robot’s intelligence consists of two facts:one is perceiving environment by a variety of sensors; the other is taking action base on the environment. In order to catch object accurately, the research of intelligent vehicle must concern about pattern recognition and cooperation of manipulator and vehicle. Therefore, it is necessary to do theoretical research and creativity. Furthermore, in order to complete a series of tasks, image processing, pattern recognition and ART neural network need cooperate with each other.This paper focuses on the functions of automatic tracking, obstacle avoidance and remote control for the intelligent vehicle, which equips with laptop, camera and manipulator. By using the laptop, firstly, processing the image that is captured by camera with the OpenCV Library. Secondly, input image’s 5*5 feature matrix to the ART neural network, recognizing whether it is the object for catching or not. Finally, transmit all control parameters to MCU and control manipulator to catch the object.In the experiment, this recognition system can really distinguish different objects, and the manipulator can catch the object successfully.